Application ASI, Three Layers
Clay Norris@claynorris
−6.7%
since Oct 1
The idea
I believe AI is crossing a line in certain domains, becoming so good that adding humans back only hurts. This basket owns that idea in layers: a core of companies where domain ASI already is the product, like recommendation and ad engines and the chip-layout duopoly, a satellite in verticals poised to cross next such as drug discovery and physical-world automation, and a light ballast of infrastructure picks kept small since that trade is already crowded.
What it owns
- Recommendation algorithms and ad auctions that demonstrate domain-specific artificial intelligence, where the system's understanding of user behavior outperforms human judgment and improves with scale. · Meta PlatformsIts feed and recommendation engine demonstrates system-level understanding of user preferences that drives engagement and monetization better than human curation., AlphabetIt operates foundational research labs and the YouTube recommendation engine, plus controls the search advertising auction where algorithmic optimization is demonstrable., AppLovinIts Axon platform optimizes real-time ad bidding at speeds and scales that outperform human media buyers in measurable outcomes., NetflixIts recommendation system drives viewing patterns for hundreds of millions of subscribers, with retention driven by algorithmic understanding rather than human programming., The Trade DeskIts Kokai platform executes real-time ad auctions at scales where artificial intelligence pricing outpaces human buyers and takes market share from manual buying.32%
- Infrastructure providers whose platforms and hardware support the training and deployment of specialized artificial intelligence across multiple domains. · NVIDIAIt supplies the primary hardware platform for training large-scale artificial intelligence systems across all specialized domains in this take., BroadcomIt manufactures custom silicon for the largest artificial intelligence infrastructure builders, providing specialized compute for training and deployment., SnowflakeIt provides the cloud platform where companies manage and prepare the data required to train artificial intelligence systems across specialized domains.19%
- Design automation tools where artificial intelligence outperforms human engineers on layout and optimization tasks, now the standard for advanced semiconductor manufacturing. · SynopsysIts design automation tools use artificial intelligence to optimize chip layouts on metrics like power and performance where human teams reach diminishing returns., Cadence Design SystemsIts place-and-route automation replaces human chip designers with artificial intelligence on tasks where the system produces results humans cannot fully explain or reproduce.14%
- Platforms combining simulation, data generation, and machine learning to accelerate drug candidate identification and validation in a domain with quantifiable objectives and measurable feedback loops. · SchrodingerIt simulates molecular interactions to rank drug candidates before laboratory testing, a bounded problem with quantifiable outcomes and rapid feedback., Tempus AIIt aggregates clinical and molecular data that trains artificial intelligence models for drug discovery, then monetizes those models to pharmaceutical partners., Recursion PharmaceuticalsIt operates automated laboratory systems that generate experimental results fed directly back into artificial intelligence models, creating short feedback loops for drug discovery., AbCellera BiologicsIt searches antibody space using artificial intelligence and automated laboratory systems at speeds and scales no human team can achieve, then partners discoveries to larger firms.14%
- Autonomous systems in logistics, agriculture, and transportation where artificial intelligence coordination replaces human operators in bounded environments with clear performance metrics. · SymboticIt deploys artificial intelligence orchestration across warehouse robotics, where the system coordinates material flow and labor more efficiently than human dispatch., Deere & CoIt develops autonomous guidance and see-and-spray systems for tractors where artificial intelligence makes real-time decisions on yield per acre based on field conditions., Aurora InnovationIt operates autonomous trucks on highways, testing whether removing the human driver improves safety and economics in a high-stakes physical environment.13%
- Cash8%